Skip to main content

Analytical Chemistry

Ammonia, Electrons, and Metals

Let’s do some pure chemistry today, because an interesting paper has come out about a reaction that every student learns about in their sophomore organic chemistry course: the Birch reduction. It’s a powerful technique that will do some things that very few other reactions will do for you (such as break up the aromaticity of benzene derivatives). That strength comes from the nature of the reagent itself. It’s hard to get more hard core about adding electrons to a compound than soaking it in a bath of solvated electrons themselves.

I’ve written about that one a few times over the years, mainly in terms of how to avoid running it under the classic conditions. Those are a mild pain in the rear to run, because you need to round up a tank of ammonia and a dry-ice condenser to rig on top of your reaction flask. You’ll be opening up that tank gradually to run a stream of ammonia gas across the cold finger, whereupon it will start dripping clear liquid ammonia (boiling point -33 degrees C) into a receiving flask. Unfortunately, if you want the best yields and reproducibility, you’re going to want to distill that first run of ammonia into your actual reaction flask (details here in a very good guide to running these reactions). There’s often some water in the system, and there can be traces of iron from the inside of the cylinder, which can mess up a Birch at catalytic levels. I have not always taken the trouble to do this, but then, my Birch reductions have not always gone as well as I would have wanted them to, either, so there is that.

Ammonia’s a pretty good solvent, a lot of organic compounds dissolve in it just fine. Birch reactions are generally run with an alcohol co-solvent (you need a proton source), and that helps even more. Once you have your starting material dissolved in this mixture, down in your dry ice bath, you start the enjoyable part: flinging in small bits of metal. The reaction can be run with lithium, sodium, potassium, and even calcium – actually figuring out which of those will work best is not so easy to do from first principles, which can at times be another complication. Most people use lithium or sodium, because they’re a lot easier to handle than potassium, from a not-producing-beautiful-magenta-flames standpoint, and straight calcium metal is relatively rare on the lab shelves.

As those bits of reactive metal go in, they dissolve in swirls of deep, vivid blue that trace through the clear ammonia solution and disappear. Here’s a video at YouTube showing just that. The blue is the color of the solvated electrons themselves, and it disappears as they react with your starting material. The classic way to tell when your reaction is done is persistence of the blue color (nothing left to react with!) But there’s an interesting phenomenon that you can see if you go wild and keep flinging metal into such a solution: the blue gets deeper and darker, but then you start making something new: a bronze-colored solution, which can be a separate layer (with the lithium reactions, I’ve seen it floating around on top of the blue phase). You can turn the whole thing to bronze phase with enough metal. But what is that stuff, and why does it look and act like that?

Well, it looks like a metal – a liquid metal. And it turns out to be an excellent conductor, too, far better than the blue solution. Now there’s a new paper that went to the trouble of doing photoelectron spectroscopy on the stuff to see what’s really going on. (Here’s some interesting background on the PhD student involved, Ryan McMullen). This is not such an easy experiment to do. The apparatus used cold microjets of the ammonia solutions spraying across a synchrotron-derived X-ray beam, which is a setup that I will assume took just a tiny bit of troubleshooting along the way. Basically, you’re using that vicious X-ray beam to blast electrons out of the surface of a sample when they absorb all that excess energy. The energy that they have as they escape depends on a lot of important characteristics – their original electronic state/energy level, for sure, as well as their rotational and vibrational state. The peaks you get are quite diagnostic and the technique is also extremely sensitive (and earned its developer, Kai Siegbahn, a Nobel Prize in 1981.

What you see when you blast the blue stuff is a peak at the “vertical displacement energy” (named for how these things are straight-up lines on an energy diagram) of about 2 electron volts. That is from the solvated electrons, which at anything short of extreme dilution are spin-paired dielectrons hanging out together. That peak is proportional to the concentration of alkali metal, and is pretty much the same no matter which metal you use, both of which point to them just being the electrons without any participation from the metals themselves.

And as you keep increasing the amount of alkali metal, that peak changes. It shades into something asymmetric, with a sharp cutoff on the low-energy end. That is a Fermi edge, and it’s just what you see with a bulk metal. I’m sure if you showed that photoelectron spectrum to someone in the field without telling them what it was, that’s exactly what they would guess as soon as they saw the trace. You also get a couple of “plasmon peaks”, which are in the visual range and account for the shiny bronze color – the same thing that accounts for the appearance of bulk metal samples as well. You can think of metals as being a “free electron gas” floating around through the metal atom lattice, and that’s what is being seen in these bronze-phase sample as well.

The gradual transition of the peaks means that the two situations (dielectrons surrounded by ammonia molecules, versus conducting delocalized electron gas) must exist at the same time and that the transition between them is smooth and not a sudden jump. No one knows yet just how that’s done – if there are tiny domains where each one of these obtains, separate from the other, or what. Those electron pairs must sort of coalesce as the concentration goes up, and you can imagine the sort of microdroplet mixture process that’s seen in some liquid-liquid phase transitions. So well before you start seeing bronze stuff in your Birch solution, you’ve already got a different situation developing down in the flask.

This work may seem pretty esoteric, and it is, in a way. But I think it’s representative of a lot of 21st-century science, in that we’re getting down to gritty microdetails about what’s really happening in physical samples and systems. You can watch crystals form in a concentrated solution, for example, but just how exactly do those molecules come together and line up so well? Do individual ones just ping around until they hit a spot they like, or do they come together as groups (how small, how large?) and assemble as larger building blocks? You can tell, similarly, that calcium ions move back and forth across living cell membranes – but just how exactly do they do that? What proteins do it, and what are their features that allow this to happen – is there a kind of “river” of ions flowing down the middle of such a protein channel, or do individual ones sort of Tarzan along from one favorable protein interaction to another? We’re really getting down to the nanoscale details, where chemistry, physics, and biology can all start to run together. Things are different down there, and we need to know what those differences are, why they exist, and how we can maneuver them to do useful things for us.

29 comments on “Ammonia, Electrons, and Metals”

  1. electrochemist says:

    Fascinating stuff! Naive question: how different are spin-paired dielectrons in an ammonia solution from Cooper pairs in a superconductor? And, is there any evidence that a substantial population of the paired dielectrons persist in the bronze solution?

    1. Ryan S. McMullen says:

      This is an excellent question. In our experiment, mentioned above, we started with dielectrons, because the solvated electron peak which would be there at dilute concentrations is buried under noise. Think of the photoelectron peaks corresponding to the ammonia molecular orbitals which were also measured simultaneously, in particular the HOMO. This peak has a huge intensity in comparison to any peak associated with excess electrons. The tails of these Gaussian peaks will swamp out very dilute spectra of solvated electrons. Not that it cannot be done, but these experiments are ongoing, but just not at the synchrotron BESSY II and will involve laser based techniques.

      I’m particularly fascinated by the dielectron photoelectron spectrum, because as you point out it is rare that electrons pair up outside of the potential well of atoms or molecules. There is still much to be done in this area.

      1. albegadeep says:

        Derek, you know you’re doing well when the people RUNNING THE EXPERIMENT apparently read your blog!

        Ryan, this is really interesting. Good work!

    2. Ben says:

      The relationship to Cooper pairing is probably fairly tenuous: the dielectrons are paired in real space, while Cooper pairs are momentum correlated in a bulk medium. I wonder if it is more fruitful to think about them as being vaguely akin to color-center impurities in crystals (e.g. NV centers in diamond), except that the vacancy is self-supporting in a liquid rather than frozen into a crystal lattice?

      I also wonder if thermodynamic (e.g. specific heat, speed of sound, tensile strength, …) measurements could/have shed more light on the phase transition from dielectron solution to metallic electron gas?

      1. Mattf says:

        Note that the momenta of electrons in Cooper pairs have equal magnitude and -opposite- directions, meaning that the paired electrons aren’t actually near each other in real space for any significant length of time. This is a classic physics oral exam question– since the electrons in a Cooper pair are fleeing from each other, in what sense are paired?

  2. Cato says:

    this reminded me of a paper in ACIE from a few years ago… A Non‐Exploding Alkali Metal Drop on Water: From Blue Solvated Electrons to Bursting Molten Hydroxide

    Check out the video in the SI!

    1. loupgarous says:

      Thanks for the link to the paper, and its supporting information (especially the video, which I just showed my wife, who’s more of an astronomer than I am a chemist, but also found it cool). Actual visual-range imagery of electrons in solution is something every chemistry class ought to show its students!

  3. GJD says:

    Birch reductions are fascinating chemistry.
    They can be a little overshadowed by the Ammonia cylinder head, which you valiantly struggled to open, refuses to close, raising all sorts of alarms!

  4. Don Monroe says:

    Seems like the YouTube link is to the photoelectron work rather than to the blue swirls.

  5. loupgarous says:

    Now my bucket list has a new entry… watching a Birch reaction, or perhaps a run of that reaction in which a solution of ruthenium and ascorbic acid are excited by a green laser to break even tightly-bound organic-halogen compounds up into less toxic things.

    Of couse, the local law enforcement guys are sure to attach a whole other motivation to “fun with great big cylinders of ammonia”. Damn meth cooks, anyway.

  6. Alan Goldhammer says:

    Good story on the Mr. McMullen and the professor who took a chance with him to seek funding for the work!!!

  7. Barry says:

    It’s been a while since we updated Pauling’s “Nature of the Chemical Bond”

  8. Rich says:

    > straight calcium metal is relatively rare on the lab shelves

    I remember we had it in high school. How else are you going to show 13 year olds the reactivity of metals?

    (Nowadays, a video, probably, any substance more energetic than sand being considered unsafe).

  9. LdaQuirm says:

    Is this the study that thunderfoot was involved in? In seems very familiar.

    1. Leak says:

      Yeah, that’s exactly that study – in fact the “video at YouTube” link is an excerpt of his longer video on the topic:

      and the link “a new paper” actually lists him (aka Philip E. Mason) as the second author:

  10. KtheKnight says:

    My first reaction during my master thesis was to repeat a Birch reduction according to the procedure in a Ph. D. thesis of a former graduate student, who got a good job in industry. I carefully cut and dissolved the sodium in the liquid ammonia and was astonished to get this bronze-colored solution instead of the expected blue solution as described in text books.
    Needless to say I got a very low yield of my organic product. I checked the reference for this reaction and … there was a typo in the thesis, it should have been 1.5 l ammonia, not 1.5 ml ammonia as written in the thesis. I did not tell anybody till now…

  11. Cf: James L. Dye’s “electrides.” A suitable metal cation in a tight crown or cryptand will crystallize with literal electrons as conter anions into bronzy well-ordered 3-D periodic lattices. What could be nicer than a sideshow of alkali metal anions?

  12. 10 Fingers says:

    These two colors (bronze and blue) were near and dear to my heart many years ago – but I saw them in different phases of material.

    I was exploring reduction of vicinal dihalides to olefins in strained molecules, and was looking at many different conditions and reducing agents for solution and gas phase chemistry.

    The “perfect blue” was a gas formed in my reaction chamber by the cloud of potassium atoms warmed under vacuum. When it was just right, the compound in a stream of argon was introduced from one end, to be trapped on a KI surface at 4K at the other for spectroscopic studies. When done right (K just a little in excess), the argon matrix itself was a most wonderful color.

    The remarkable, unique bronze color was that of freshly made potassium in graphite. I always thought of it as “powdered electrons.” If it didn’t make my heart skip a beat at the beauty of it when I had made it, then it wasn’t a good prep.

    Short only of the color of the crimson flame torching from the end of a column of solid methyllithium on glass helices (just from blowing off the excess) these were at the top of my “favorite colors of things I worked with” list.

    1. Derek Lowe says:

      Alkyllithium flames can indeed be really nice – when you’re expecting them, anyway.

      1. Orange_Pills says:

        Dropping a bit of t-BuLi on a mixture of isopropanol and water has been a beautiful sight. It’s educational too! Should be included in more programs.

  13. gippgig says:

    Does the bronze solution behave differently if you try to do a Birch reduction with it (yes, I realize you’d have to add the starting material after rather than before the metal)?

    1. Derek Lowe says:

      It doesn’t work nearly as well, is what I’ve always heard.

  14. photoelectron fan says:

    Really cool and challenging experiment. Thanks for writing about it!

  15. Paul W. says:

    You mention that calcium is effective in Birch reductions. If so, wouldn’t strontium or barium be even more so, in the same way that potassium is more effective than lithium or sodium (but more dangerous, so less employed)?

  16. li zhi says:

    Hmmm. A nearly random association. Back in the 70’s I was looking at graphite intercalation with SbFx and saw both blue and bronze colored materials. We were looking for practical uses for our Sb2O5 – and high conductivity of the intercalate had been observed. High water sensitivity made real world use impractical.

  17. Tom A says:

    Oy! Times have changed since I studied upper level organic (Morrison and Boyd) back in the early seventies. If the Birch reaction is in that text I must have dozed off in the lecture.
    Thanks for the respite from Covid-19.

  18. kultakutri says:

    I don’t have much idea about what’s going on in the paper beyond prepositions but Pavel Jungwirth does a good job popularising science and writes cute little op-eds.

  19. David Edwards says:

    Aside from TIWWW, it’s posts like this I keep coming back to read, because Derek introduces me via said posts, to a wealth of fascinating new insights into phenomena that were considered “done and dusted” in earlier textbooks.

    At this point, however, I do have a question – when was the concept of ‘solvated electrons’ first postulated, and considered to be applicable to this reaction?

  20. Ed C says:

    While this is somewhat off-topic, but has anyone tried running Birch reductions under pressure? Ammonia’s vapor pressure at 20 Celsius is only about 9.5 bar, which is pretty low.

Comments are closed.